Table 2. The rate of change in PIPs in the intervention group versus the control group after adjusting for the GP subgroup characteristics. Estimates of IRRs showing the changes in PIPs per 100 prescriptions at post-intervention relative to 1 year pre-intervention obtained from the Poisson cluster effects regression model.


	IRR (95% CI)	P-value
PIPs at baseline (ref: control group)
Intervention	1.10 (0.97 to 1.25)	0.15
Within-group reduction of PIPs relative to baseline
Control	0.93 (0.90 to 0.96)	<0.01
Intervention	0.80 (0.78 to 0.82)	<0.01
^aReduction of PIPs after intervention (ref: control group)
Intervention	0.86 (0.83 to 0.90)	<0.01
Subgroups:
Age group, years (ref: 27–42 years)
43–48	0.85 (0.81 to 0.89)	<0.01
49–52	0.92 (0.87 to 0.97)	<0.01
53–56	1.00 (0.95 to 1.05)	0.94
57–68	1.21 (1.16 to 1.27)	<0.01
Specialist (ref: non-specialist)
Specialist	1.04 (0.99 to 1.08)	0.13
Type of practice (ref: single-handed)
Group practice	1.04 (0.99 to 1.08)	0.07
Sex (ref: female)
Male	1.40 (1.35 to 1.44)	<0.01
Practice setting (ref: rural)
Urban	1.15 (1.11 to 1.19)	<0.01
Mean PIPs p 100 prescriptions at baseline (ref: ≤1.6)
1.7–2.0	1.40 (1.34 to 1.45)	<0.01
2.1–2.3	1.77 (1.71 to 1.84)	<0.01
2.4–2.9	2.10 (2.02 to 2.18)	<0.01
>2.9	2.40 (2.30 to 2.50)	<0.01
Prescriptions per patient at baseline (ref: <8.7)
8.7–10.7	1.44 (1.38 to 1.51)	<0.01
10.8–12.8	1.96 (1.88 to 2.05)	<0.01
12.9–14.9	2.31 (2.20 to 2.42)	<0.01
≥15	2.97 (2.83 to 3.12)	<0.01

IRR = incidence rate ratio. PIP = potentially inappropriate prescription. Ref = reference. ^aPIPs were 14% (1.0 to 0.86) lower in the intervention arm compared to the control at after intervention.